GUIDES / AUTOMATION

Workflow automation: what to automate and how to keep it running.

Workflow automation moves repeated work, such as copying orders, updating stock or sending follow-ups, into software that follows fixed rules. This page covers which tasks are worth it and what keeps a workflow running once it is live.

Automation serviceConsulting by the hour

WHAT THIS AREA COVERS

Automation works when the task, its exceptions and the person in charge are clear.

Most automation trouble has little to do with the tool. It comes from a workflow nobody looks after, exceptions that were never written down and failures that go unnoticed for weeks, because nobody can explain the workflow end to end. That is where this area starts; comparing tools comes later.

It is written for teams that copy data between tools, chase the same approvals or rebuild the same report every week. General business tasks such as leads, invoices and reports sit here; orders and stock moving between sales channels have their own page.

  • Tasks worth automating, and tasks better left manual
  • Tool choice as the last decision, once the work is written down
  • Alerts and checks so that a stopped workflow gets noticed
Abstract composition of a task passing between tools

READ NEXT

Pages that take each part further.

Business automation

Workflow design, tool choice, exception handling and build-or-buy decisions for owners and operations leads.

Read the topic →

Multichannel ecommerce

Orders, stock and prices moving between your own shop, marketplaces and the other channels you sell through.

Read the topic →

AI systems guides

For tasks that need judgement instead of fixed rules, including a guide on the order for bringing AI into operations.

See the guides →

GUIDES

Automation guides.

Business automation, in more depth

That page compares n8n, Make, Zapier and custom code and explains how exceptions reach a person. The basics are in the questions below.

FREQUENTLY ASKED

Questions about workflow automation.

What is the difference between automation and AI?
Automation follows fixed rules, so the same input always gives the same output. AI handles inputs that vary and produces output that depends on context. A workflow that calls a language model in one step is still automation; when it stops working without the model, it has become an AI system.
Which tasks should not be automated?
Tasks whose process changes every month, tasks nobody is responsible for, and manual steps that exist to catch errors. Automating them moves the problem into software, where it is harder to see.
Who looks after an automation once it runs?
A named person on your team who receives the failure alerts and decides on exceptions. Without one, a stopped workflow can sit unnoticed until someone misses its output.
How do you know an automation is healthy?
Track how often it completes, how many cases fall out as exceptions, how often someone overrides it and how long a failure takes to be noticed. A workflow that never reports an exception may be skipping cases it should have flagged.
What if the tool changes its pricing or drops a feature?
Critical workflows are documented and kept portable enough to move to another tool at a known cost. For minor workflows, accepting the dependency is usually cheaper than guarding against it.

SERVICES

When the workflow is ready to build.

Automation

A quote based on the monthly hours the workflow saves, with the software delivered under a licence of use.

See service →

AI agents

For the one step inside a workflow that needs judgement, with a narrow scope and a person reviewing it.

See service →

Consulting by the hour

A paid session to decide whether a task should be automated, handled by a tool you buy or left as it is.

See service →

Tell us what needs doing.

Amazon work, a question for a consulting session or a process worth automating. We reply with the service that fits.

Contact usSee the services